FlexIO : Location-flexible Execution of In Situ Data Analytics for Large Scale Scientific Applications

Increasingly severe I/O bottlenecks on High-End Computing machines are prompting scientists to proc ess simulation output data while simulations are running and before placing data on disk – ”in situ” and/or ”in-transit ”. There are several options in placing in-situ data analytics a long the I/O path: on compute nodes, on staging nodes dedicated to analytics, or after data is stored on persistent storage. Diff erent placements have different impact on end to end performance and cost. The consequence is a need for flexibility in the locati on of in situ data analytics. The FlexIO facility described in this paper supports flexible placement of in situ analytics, by offering simple abstractions and methods that help developers explo it the opportunities and trade-offs in performing analytics at different levels of the I/O hierarchy. Experimental results with several large-scale scientific applications demonstrate the importance of flexibility in analytics placement. Keywords-I/O, In Situ Processing, Staging, Placement, Data Analytics

[1]  Jarek Nieplocha,et al.  Evaluation of active storage strategies for the lustre parallel file system , 2007, Proceedings of the 2007 ACM/IEEE Conference on Supercomputing (SC '07).

[2]  Kenneth Moreland,et al.  Sandia National Laboratories , 2000 .

[3]  Anthony Mezzacappa,et al.  Petascale supernova simulation with CHIMERA , 2007 .

[4]  Emmanuel Jeannot,et al.  Adaptive online data compression , 2002, Proceedings 11th IEEE International Symposium on High Performance Distributed Computing.

[5]  David R. O'Hallaron,et al.  Scalable systems software - From mesh generation to scientific visualization: an end-to-end approach to parallel supercomputing , 2006, SC.

[6]  Karsten Schwan,et al.  Six degrees of scientific data: reading patterns for extreme scale science IO , 2011, HPDC '11.

[7]  Margaret H. Wright,et al.  The opportunities and challenges of exascale computing , 2010 .

[8]  Michael E. Papka,et al.  Toward simulation-time data analysis and I/O acceleration on leadership-class systems , 2011, 2011 IEEE Symposium on Large Data Analysis and Visualization.

[9]  Fei Meng,et al.  Functional Partitioning to Optimize End-to-End Performance on Many-core Architectures , 2010, 2010 ACM/IEEE International Conference for High Performance Computing, Networking, Storage and Analysis.

[10]  Scott Klasky,et al.  Examples of in transit visualization , 2011, PDAC '11.

[11]  Mahadev Satyanarayanan,et al.  Diamond: A Storage Architecture for Early Discard in Interactive Search , 2004, FAST.

[12]  Ron Oldfield,et al.  Improving Data Access for Computational Grid Applications , 2003, Cluster Computing.

[13]  Jeffrey Scott Vitter,et al.  Distributed computing with load-managed active storage , 2002, Proceedings 11th IEEE International Symposium on High Performance Distributed Computing.

[14]  P. Balaji,et al.  GePSeA: A General-Purpose Software Acceleration Framework for Lightweight Task Offloading , 2009, 2009 International Conference on Parallel Processing.

[15]  J. Manickam,et al.  Gyro-kinetic simulation of global turbulent transport properties in tokamak experiments , 2006 .

[16]  Margo I. Seltzer,et al.  Network-Aware Operator Placement for Stream-Processing Systems , 2006, 22nd International Conference on Data Engineering (ICDE'06).

[17]  Gregory R. Ganger,et al.  Dynamic Function Placement for Data-Intensive Cluster Computing , 2000, USENIX Annual Technical Conference, General Track.

[18]  Prabhat,et al.  FastBit: interactively searching massive data , 2009 .

[19]  Christian H. Bischof,et al.  VIRACOCHA: An Efficient Parallelization Framework for Large-Scale CFD Post-Processing in Virtual Environments , 2004, Proceedings of the ACM/IEEE SC2004 Conference.

[20]  Karsten Schwan,et al.  A Type System for High Performance Communication and Computation , 2011, 2011 IEEE Seventh International Conference on e-Science Workshops.

[21]  B. Fryxell,et al.  FLASH: An Adaptive Mesh Hydrodynamics Code for Modeling Astrophysical Thermonuclear Flashes , 2000 .

[22]  Karsten Schwan,et al.  Event-based systems: opportunities and challenges at exascale , 2009, DEBS '09.

[23]  Seetharami R. Seelam,et al.  Modeling the Impact of Checkpoints on Next-Generation Systems , 2007, 24th IEEE Conference on Mass Storage Systems and Technologies (MSST 2007).

[24]  Kun-Lung Wu,et al.  COLA: Optimizing Stream Processing Applications via Graph Partitioning , 2009, Middleware.

[25]  Edward A. Lee,et al.  Scientific workflow management and the Kepler system , 2006, Concurr. Comput. Pract. Exp..

[26]  Jacqueline H. Chen,et al.  Direct numerical simulation of turbulent combustion: fundamental insights towards predictive models , 2005 .

[27]  Daniel S. Katz,et al.  Pegasus: A framework for mapping complex scientific workflows onto distributed systems , 2005, Sci. Program..

[28]  Karsten Schwan,et al.  Just in time: adding value to the IO pipelines of high performance applications with JITStaging , 2011, HPDC '11.

[29]  Ray W. Grout,et al.  Ultrascale Visualization In Situ Visualization for Large-Scale Combustion Simulations , 2010 .

[30]  John Giacomoni,et al.  FastForward for efficient pipeline parallelism: a cache-optimized concurrent lock-free queue , 2008, PPoPP.

[31]  Karsten Schwan,et al.  In-situ I/O processing: a case for location flexibility , 2011, PDSW '11.

[32]  Karsten Schwan,et al.  XChange: coupling parallel applications in a dynamic environment , 2004, 2004 IEEE International Conference on Cluster Computing (IEEE Cat. No.04EX935).

[33]  Choong-Seock Chang,et al.  Spontaneous rotation sources in a quiescent tokamak edge plasma , 2008 .

[34]  Michael E. Papka,et al.  In situ data analysis and I / O acceleration of FLASH astrophysics simulation on leadership-class system using GLEAN , 2011 .

[35]  Karsten Schwan,et al.  Managing Variability in the IO Performance of Petascale Storage Systems , 2010, 2010 ACM/IEEE International Conference for High Performance Computing, Networking, Storage and Analysis.

[36]  R. Samtaney,et al.  Grid -Based Parallel Data Streaming implemented for the Gyrokinetic Toroidal Code , 2003, ACM/IEEE SC 2003 Conference (SC'03).

[37]  Karsten Schwan,et al.  PreDatA – preparatory data analytics on peta-scale machines , 2010, 2010 IEEE International Symposium on Parallel & Distributed Processing (IPDPS).

[38]  Ron Oldfield,et al.  Extending scalability of collective IO through nessie and staging , 2011, PDSW '11.